Bain & Company8 Sep 2026Profit shift
AI will move $4.7 trillion of profit by 2035, and retail logistics must defend, not just adopt
Bain analysed 92 sectors to map how AI will redistribute $4.7 trillion of corporate profit through 2035, more than three times the internet's impact, in roughly half the time. Automotive, logistics and freight sit in Bain's highest-exposure cluster, where AI erodes existing advantage, while retail's near-term ceiling looks more like dynamic pricing and demand forecasting.
Why it matters
Treat the $4.7 trillion as Bain's modelling, not a measured outcome. The practical distinction is defend versus adopt: in logistics and freight, standing still is a competitive loss, while for most retailers AI's current ceiling is pricing and forecasting rather than a rebuilt business model.
EY15 Sep 2026Value gap
EY: consumer products CEOs are raising AI investment, but only 12% say its impact reaches senior management review
EY surveyed more than 850 senior consumer products executives across 24 markets in early 2026. 73% of CEOs raised planned AI investment and 37% report measurable impact in supply chain and procurement, but just 12% say that impact is tied to financial reporting and regularly reviewed by senior management.
Why it matters
The gap between 73% spending more and 12% tracking the result in the numbers leadership reviews is where AI value goes unproven. Ask any AI programme which of those three figures its own reporting would put it in.
McKinsey Quarterly9 Sep 2026Operating model
Firms that redesign work around AI outperform those that just add tools
A survey of more than 700 executives splits organisations into three horizons: enablement, automation and reinvention. Only 13% count as reinventors, but 48% of them report meaningful enterprise value against 24% for automation and 13% for enablement, and they report faster decision cycles than firms still adding tools to unchanged processes.
Why it matters
The finding gives a concrete diagnostic: which horizon is an organisation in, and does its reported value match what that horizon should deliver. Reinvention isn't a single copyable template; McKinsey found reinventors differ in which parts of their operating model they choose to redesign.
McKinsey Digital11 Sep 2026Cybersecurity
The gap between a vulnerability going public and being exploited has fallen from weeks to hours
McKinsey cites industry tracking showing the average time between a critical vulnerability's disclosure and active exploitation has fallen to a matter of hours, down from roughly three weeks in 2025, as frontier AI models can now generate working exploits at scale. Its diagnosis is organisational: no single function owns decision speed across IT, legal, procurement and security.
Why it matters
This reframes AI cyber risk as a governance problem rather than a tooling one. McKinsey's proposed starting point, a defined 'minimum viable organisation' (a small core whose disruption would be existential), is a useful frame for deciding which systems need continuous monitoring first.
Gartner9 Sep 2026Workforce
30% of AI-driven layoffs will need to be rehired by 2029, at higher cost
Gartner's Hype Cycle for the Future of Work predicts that by 2029, 30% of employees laid off due to AI replacement will need to be rehired, often at a significantly higher cost, because workforce cuts made for short-term financial gain erode institutional knowledge and talent pipelines.
Why it matters
This puts a hard number on a trade-off many AI business cases don't price in: the cost of reversing a workforce cut once a capability gap becomes obvious. Any headcount reduction funding an AI programme deserves a second look at what rehiring would cost if the cut proves premature.
MIT Sloan Management Review / BCG8 Sep 2026Governance
72% of an expert panel say treating AI agents as accountable decision-makers is a governance failure
A panel of 29 international AI experts convened jointly by MIT Sloan Management Review and BCG rated agreement with the idea that governance treating agents as autonomous decision-makers will fail; 72% agreed. Their argument is that an agent's operational autonomy doesn't create legal or moral responsibility, so accountability has to sit with a named human.
Why it matters
This is expert consensus, not measured behaviour, so treat 72% as informed opinion rather than a finding about what companies are doing. Its practical use is a governance checklist: for any agent deployed into stock, pricing or replenishment decisions, someone specific should be named accountable before a bad decision happens, not after.
